academic-source-analyzer
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructions require the execution of a specific local Python script (
.codex/skills/scripts/extract_pdf_smart.py) viauv runto process PDF files. This execution is scoped to a specific internal script path. - [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted academic source materials provided by the user, which presents a surface for instructions embedded in those documents to influence agent behavior.
- Ingestion points: Academic source material and PDF files referenced in
SKILL.md. - Boundary markers: No specific delimiters or boundary markers are defined for the input source material.
- Capability inventory: The agent can execute a local PDF extraction script (
extract_pdf_smart.py) and is instructed to interact with anacademic-reference-findertool. - Sanitization: The skill explicitly instructs the agent to screen for credentials, private data, respondent identities, and confidential material as the first step of the process.
Audit Metadata